Hello everybody, I am a graduate student from South China Normal University, and my research field is spatial analysis and geosimulation.
Since 2010-11-15 when I sent the 1st e-mail to Prof. Clarke, my experience using SLEUTH to provide decision support to Dianchi watershed formally started. With the precious advices from Prof. Clarke, his student Guofeng Cao, Gargi Chaudhuri, and some experienced SLEUTHers inChina, the problems I met are solved one by one. Now I’d like to summarize my experience using SLEUTH during different phases as follows.
(1) SLEUTH can be complied under UNIX and Linux, and Cygwin (Cygwin is a Unix/ Linux environment under Windows XP). SLEUTH support parallel computing unless MPICH is installed, and it’s quite helpful when the dataset used is large.(e.g. my study area has a extent of 1748 columns*3688 rows, and each cell is30m*30m, it needs parallel computing )
(2) Gargi Chaudhuri told me that “The data format and pixel values becomes an issue if you go back and forth between different software esp. with ArcGIS”. So I use ArcGIS alone to prepare the input gif dataset. Make sure that all data have been registered to the same geographic projection with the same extent, resolution, and the naming convention.
Data flow under ArcGIS: GRIDIMAGE -> TIF->GIF (pixel depth: unsigned 8 bit)
Naming convention → http://www.ncgia.ucsb.edu/projects/gig/Imp/imSetUp.htm.
(3) I downloaded SLEUTH3.0beta_p01_linux.tar.gz, followed the instructions as http://www.ncgia.ucsb.edu/projects/gig/Imp/imVerify.htm has mentioned, and compiled it successfully under Cygwin and Linux (both Ubuntu Linux and SUSE Linux). Then I followed the user’s guide of MPICH2 to set up the parallel computing environment for SLEUTH. → MPICH2 and the user’s guide is available at http://phase.hpcc.jp/mirrors/mpi/mpich2/
(4) SLEUTH is suggested to be calibrated under the same resolution during coarse, fine and final calibration, and make sure there are four urban time periods during calibration (or the model will simulate linear growth).
Traditionally, Lee-Salee Metric is used to selecting coefficient ranges. The optimum SLEUTH Metric (OSM) is widely applied after DietZel and Clarke proposed it in 2007. Actually, to get the best coefficient sets for different research areas, each range selection method may be different based on the idea of single metric or composite metric.
→ Dietzel C, Clarke K C. Towards optimal calibration of the SLEUTH land use change model. Transactions in GIS, 2007, 11(1): 29-45.
(5) SLEUTH has strong ability to make forecasts, and integrates well with scenario-based planning. Clarke has once summarized that there are 5 methods to generate scenarios (Clarke, 2008). In this process, special attention should be paid to the value settings in the exclusion layer. The value should be given between (0, 100) according to the possibility of urban development from high to low.
→Clarke K C. A Decade of Cellular Urban Modelling with SLEUTH: Unresolved issues and Problems. Planning Support Systems for Cities and Regions, 2008: 47-60.
(6) Spatial metrics shows strong ability to interpret and evaluate the modelling results. Noted that the output gif files need registration when you import them into Fragstats. Meanwhile, spatial analysis methods like map algebra, buffer analysis .etc may also be good choice.
→ Herold M, Goldstein N C, Clarke K C. The spatiotemporal form of urban growth: measurement, analysis and modeling. Remote Sensing of Environment, 2003, 86(3): 286-302
Herold M, Couclelis H, Clarke K C, The role of spatial metrics in the analysis and modeling of urban land use change. Computers, Environment and Urban Systems,2005,29(4): 369–399.